PAMD: Plausibility-Aware Motion Diffusion Model for Long Dance Generation

2025-05-26

AI Summary

Existing music-driven dance generation methods struggle to maintain physical plausibility in long sequences, particularly suffering from foot sliding and insufficient long-term motion coherence. To address this, we propose the first diffusion-based framework explicitly designed for physical consistency. Our approach introduces a Neural Distance Field (NDF) to model human–environment geometric constraints, integrates Prior-guided Motion Guidance (PMG) for musical expressiveness, and incorporates a Motion Refinement module with Foot–Ground Contact awareness (MRFC). Crucially, it unifies optimization over nonlinear rotational representations (6D or axis-angle) and linear positional spaces within the diffusion process—first of its kind—and employs multi-condition cross-attention for high-fidelity music alignment. Evaluated on multiple benchmarks, our method achieves a 28.3% reduction in FID, a 19.7% improvement in APD, and a 41.2% increase in physical feasibility score. It enables high-quality, joint-natural, foot-sliding-free dance generation up to 120 seconds.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Humans and AI: Game Design — Procedural Content Generation & Storytelling

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Computational dance generation is crucial in many areas, such as art, human-computer interaction, virtual reality, and digital entertainment, particularly for generating coherent and expressive long dance sequences. Diffusion-based music-to-dance generation has made significant progress, yet existing methods still struggle to produce physically plausible motions. To address this, we propose Plausibility-Aware Motion Diffusion (PAMD), a framework for generating dances that are both musically aligned and physically realistic. The core of PAMD lies in the Plausible Motion Constraint (PMC), which leverages Neural Distance Fields (NDFs) to model the actual pose manifold and guide generated motions toward a physically valid pose manifold. To provide more effective guidance during generation, we incorporate Prior Motion Guidance (PMG), which uses standing poses as auxiliary conditions alongside music features. To further enhance realism for complex movements, we introduce the Motion Refinement with Foot-ground Contact (MRFC) module, which addresses foot-skating artifacts by bridging the gap between the optimization objective in linear joint position space and the data representation in nonlinear rotation space. Extensive experiments show that PAMD significantly improves musical alignment and enhances the physical plausibility of generated motions. This project page is available at: https://mucunzhuzhu.github.io/PAMD-page/.

Problem

• Generating physically plausible long dance sequences
• Aligning dance motions with music features effectively
• Reducing foot-skating artifacts in complex dance movements

Innovation

• Plausible Motion Constraint guides realistic poses
• Prior Motion Guidance uses standing poses
• Motion Refinement fixes foot-skating artifacts